A&A 536, L1 (2011) DOI: 10.1051/0004-6361/201118089 & c ESO 2011 Astrophysics

Letter to the Editor

First firm spectral classification of an early-B pre-main-sequence : B275 in M 17

B. B. Ochsendorf1, L. E. Ellerbroek1, R. Chini2,3,O.E.Hartoog1,V.Hoffmeister2,L.B.F.M.Waters4,1, and L. Kaper1

1 Astronomical Institute Anton Pannekoek, University of Amsterdam, Science Park 904, PO Box 94249, 1090 GE Amsterdam, The Netherlands e-mail: [email protected]; [email protected] 2 Astronomisches Institut, Ruhr-Universität Bochum, Universitätsstrasse 150, 44780 Bochum, Germany 3 Instituto de Astronomía, Universidad Católica del Norte, Antofagasta, Chile 4 SRON, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands

Received 14 September 2011 / Accepted 25 October 2011

ABSTRACT

The optical to near- (300−2500 nm) spectrum of the candidate massive (YSO) B275, embedded in the star-forming region M 17, has been observed with X-shooter on the ESO Very Large Telescope. The spectrum includes both photospheric absorption lines and emission features (H and Ca ii triplet emission lines, 1st and 2nd overtone CO bandhead emission), as well as an infrared excess indicating the presence of a (flaring) circumstellar disk. The strongest emission lines are double-peaked with a peak separation ranging between 70 and 105 km s−1, and they provide information on the physical structure of the disk. The underlying photospheric spectrum is classified as B6−B7, which is significantly cooler than a previous estimate based on modeling of the spectral energy distribution. This discrepancy is solved by allowing for a larger stellar radius (i.e. a bloated star) and thus positioning the star above the . This constitutes the first firm spectral classification of an early-B pre-main-sequence (PMS) star. We discuss the position of B275 in the Hertzsprung-Russell diagram in terms of PMS evolution. Although the position −5 −1 is consistent with PMS tracks of heavily accreting (M˙ acc ∼> 10 M yr ), the fact that the of the object is detectable suggests that the current - rate is not very high. Key words. : formation – stars: pre-main-sequence – stars: massive – stars: variables: T Tauri, Herbig Ae/Be

1. Introduction Infrared surveys have revealed several hundred candidate massive YSOs, based on arguments (e.g., Urquhart Observational and theoretical evidence is accumulating that the et al. 2011). A (K-band) spectrum has been obtained for only formation process of massive stars is through disk accretion, afewofthese(Hanson et al. 1997, 2002; Bik et al. 2006), similar to low-mass stars. This persists despite the strong ra- and they show a red continuum, likely due to hot dust, and an diation pressure and ionizing power produced by the massive emission-line spectrum that includes Brγ and, often, CO 2.3 μm young stellar object (YSO) that may reverse the accretion flow bandhead emission. The latter emission can be modeled as being and prevent matter from accreting onto the forming star (e.g., produced by a Keplerian rotating disk surrounding the young, Keto et al. 2006; Krumholz et al. 2009). Given the short main- potentially massive star (Bik & Thi 2004; Blum et al. 2004; sequence lifetime of massive stars, the mass accretion rate must −3 −1 Wheelwright et al. 2010). be high (up to ∼10 M yr , Hosokawa et al. 2010) to ensure that the star is not leaving the main sequence before the accretion As massive stars show most spectral features in the UV and process has finished. optical ranges, the study of their photospheric properties would Evidence of accretion must come from the detection of cir- strongly benefit from extending the spectral coverage as far to cumstellar disks, and possibly bipolar jets, as observed around the blue as possible. Obviously, by the surrounding forming low-mass stars (e.g., Appenzeller & Mundt 1989). gas and dust makes this an observational challenge. Only in rare Disks and outflows around massive YSO candidates are being cases have spectra of candidate massive YSOs been obtained at reported (e.g., Chini et al. 2004; Kraus et al. 2010; Ellerbroek optical wavelengths. Hanson et al. (1997) obtained optical and et al. 2011), but the physical properties of the forming massive near-infrared spectra of candidate massive YSOs in M 17, one stars remain uncertain. The mass of the central object has to be of the most massive nearby star-forming regions in the estimated from the emerging flux, and the direct detection of the (Hoffmeister et al. 2008; Broos et al. 2007; Povich et al. 2009). photospheric spectrum turns out to be very difficult at this early For the “normal” OB stars Hanson et al. (1997) found a good stage of evolution (e.g., Testi et al. 2010). correspondence between the optical and K-band spectra, but the massive YSO optical spectra remained inconclusive. For four Based on observations performed with the ESO Very Large massive YSO candidates, they registered the optical spectrum Telescope on Cerro Paranal, Chile, as part of the X-shooter Science from 400 to 480 nm, indicating a high mass and luminosity. The Verification program 60.A-9402(A). blue spectrum of the strong CO emission source B275 showed Article published by EDP Sciences L1, page 1 of 4 A&A 536, L1 (2011) H HeI H HeI HeI DIB HeI MgII H H HeI SiII H 4 CaII CII 3.5 c B2V + 3 B3V ux

fl B5V 3.0 2 B275 H norm. B7V 1 B8V 2.5 CaII 849.8 380 400 420 440 460 (nm) CaII 854.2 2.0 1.5 norm. flux + c CaII 866.2 1.4 1.3 1.5 OI 844.6

1.2 (4-2) (5-2) (2-0) (3-0) 1.1 (3-1) (4-1) (5-3) (6-3) [OI] 630.0 1.0 1.0 0.9 0 5 10 15 -400 -200 0 200 400 Fig. 1. Top left: the blue spectrum of B275 in M 17 shown next to B main-sequence-star spectra (Gray & Corbally 2009). Bottom left:the 1st and 2nd overtone CO emission bands. Zero velocity corresponds to the first component in the series (at 2294 and 1558 nm, respectively). Right: a sample of the emission line profiles in the spectrum of B275. The Ca ii triplet lines and O i 845 nm are superposed on Paschen series absorption lines. The flux of the Hα line is scaled down by a factor 5; the structure near the peak is a remnant of the nebular-line subtraction. no definite photospheric features other than hydrogen, so that 3. Results the of this source remained uncertain. The spectral en- ergy distribution (SED), though, indicated spectral type late-O In the following we present the results for the accurate classifica- or early B, at an adopted distance of 1.3 kpc. We set out to tion of the photospheric spectrum, analyze the interstellar spec- exploit the high efficiency and broad wavelength coverage of trum to determine the extinction, model the SED using the flux- the new medium-resolution spectrograph X-shooter on the ESO calibrated X-shooter spectrum, and describe the emission-line Very Large Telescope (VLT) to (i) detect the photospheric spec- spectrum produced by the circumstellar disk. trum of B275 in M 17; (ii) determine its effective temperature in order to place the candidate massive YSO unambiguously onto 3.1. Spectral classification recent evolutionary tracks; and (iii) search for ongoing accretion activity and investigate the structure of the disk. Hydrogen absorption lines were detected by Hanson et al. (1997) in the blue spectrum of B275, but do not allow for an accu- rate spectral classification. As shown in Fig. 1, a number of 2. VLT/X-shooter observations of B275 and metal lines can be used to classify the photo- VLT/X-shooter spectra were obtained of the massive YSO B275 spheric spectrum. The He i 400.9 nm and C ii 426.7 nm, in M 17 (CEN 24, RA(2000.0) = 18h20m25s.13, Dec(2000.0) = prominent down to spectral type B3, are very weak. The −16◦1024. 56, V = 15.55 mag, K = 8.05 mag, Chinietal. He i 447.1 nm/Mg ii 448.1 nm ratio is a useful spectral indicator 1980; Skrutskie et al. 2006) on August 11, 2009 at 03h20 UT, for mid- to late-B stars (Gray & Corbally 2009) as the neutral during the first science verification run (PI Chini). The observa- helium line disappears towards lower temperature (A0) and the tions in the UVB arm (300−600 nm) were binned (2 pixels) in magnesium line strengthens. When also considering another line the wavelength direction in order to increase the signal-to-noise ratio, Si ii 412.8 nm/He i 448.1 nm, the spectral type becomes B6 ratio of this part of the spectrum, while still oversampling the (±one subtype). resolution element. The 1.6 slit was used resulting in resolving The spectral type and luminosity class of B275 are further power R = 3300. For the VIS (550−1000 nm) and the NIR arm constrained by comparison of the observed H i and He i line pro- (1000−2500 nm) a 0.9 slit was used (R = 8800 and 5600, re- files (as well as the shape of the SED, see Sect. 3.3), to model spectively). The total exposure time was 45 min, resulting in profiles produced with FASTWIND (Puls et al. 2005). This code a typical signal-to-noise ratio of 70. For more details on the calculates non-LTE line-blanketed stellar models X-shooter instrument and its performance, see D’Odorico et al. and is especially suited to modeling stars with strong winds, but (2006); Vernet et al. (2011). The observing conditions were good it can also be used to examine Teff and log g dependent pho- (0.6 seeing in V and 76% Moon illumination). The spectra were tospheric lines of H and He. We constructed a grid of models obtained by nodding the star on the slit, allowing for background (in varying Teff and log g)ofB6−B8 dwarf and giant stars. The subtraction. The standard procedures of data reduction were ap- synthetic H i and He i profiles resulting from the models are plied using the X-shooter pipeline version 0.9.4 (Goldoni et al. convolved with the corresponding instrumental and rotational −1 2006; Modigliani et al. 2010). For flux calibration and telluric profiles. We adopt vr sin i = 100 km s . An acceptable fit is absorption correction, the standard stars EG274 and HD 180699 obtained for a B7 V model (Fig. 2);however,thebestfitis were used. obtained for a B7 III model, with Teff = 13 000 ± 500 K and

L1, page 2 of 4 B. B. Ochsendorf et al.: First firm spectral classification of an early-B PMS star: B275 in M 17

1.2 1.2 B6 III B6 V B7 III B7 V 10-8 B8 III B8 V 1.0 1.0

0.8 0.8 10-9

0.6 0.6 ) -2 10-10 cm

0.4 0.4 -1 Kurucz model:

3940 3960 3980 4000 4020 3940 3960 3980 4000 4020 (erg s λ Teff = 13000 K

F -11 λ 10 log g = 3.5 B6 III B6 V

Normalized flux R = 8.1 R • B7 III B7 V  O B8 III B8 V d = 1980 pc 1.0 1.0 log (L/LO •)= 3.2 10-12 Extinction law:

AV = 6.1

0.9 0.9 RV = 3.3 10-13 1000 10000 λ (nm) 0.8 0.8 4460 4470 4480 4490 4460 4470 4480 4490 Fig. 3. The flux-calibrated X-shooter spectrum of B275 from − λ (A) 300 2500 nm (black) along with the photometric data (red triangles, black error bars) from Chini et al. (1980)(UVBRI), 2MASS (Skrutskie Fig. 2. FASTWIND model profiles of H (top)andHei 447.1 nm et al. 2006, JHK), Spitzer GLIMPSE (Benjamin et al. 2003,3.6,and (bottom) lines for B6−B8 giants (left) and main-sequence stars (right). 5.8 μm), and Nielbock et al. (2001) (N, Q). When dereddened (AV = The B7 III model provides the best fit with the observed profiles. 6.1 mag, orange line, blue diamonds), the SED is described well by a B7 III Kurucz model (blue, dashed line). The excess flux at 500−800 nm is an instrumental feature. log g = 3.5 ± 0.3. This is the first time that the spectral type of a candidate massive YSO has been accurately determined. larger than that of, e.g., a B5 zero-age main sequence (ZAMS) star (2.7 R, Hanson et al. 1997). An additional constraint is 3.2. Interstellar spectrum provided by the height of the Balmer jump, which also varies The optical spectrum of B275 includes several interstellar fea- with Teff and R:Fig.3 demonstrates that the Balmer jump tures: atomic resonance transitions (e.g., Ca ii H&K, Na i D) (as well as the Paschen jump) is nicely fit to the observed spec- and diffuse interstellar bands (DIBs). The DIB strength provides trum. Thus, with a larger radius the discrepancy between the a measure of the interstellar extinction. For the DIBs centered classification of the photospheric spectrum and the dereddened at 578.0, 579.7, and 661.4 nm, we measure an equivalent width SED is solved. The consequence is that B275 is not on the main of 0.063, 0.014, and 0.021 nm, respectively, with a typical er- sequence but is a so-called bloated star, where the appropriate ror of 10%. Using the relations from Cox et al. (2005), we ar- spectral type would be B7 III. The corresponding luminosity is rive at an E(B − V)of1.0 ± 0.1 mag. For an average value of log L/L = 3.2. RV = 3.1, these DIB strengths yield AV  3 mag of visual ex- tinction. This is less than the determination of A  6.1mag V 3.4. Accretion signatures from dereddening the SED (Sect. 3.3). Hanson et al. (1997) note that the DIB features in spectra of M 17 stars do not show large A pronounced, double-peaked emission feature is detected in the variations in strength, despite the fairly wide range in total ex- strongest H Balmer lines, the Ca ii triplet and the O i 844.6 nm tinction, from AV = 3 − 10 mag. We consider their explanation line (Fig. 1). The measured peak-to-peak separation ranges likely that the DIBs are mostly tracing the foreground dust and from 71 ± 7kms−1 (O ii 844.6 nm) to 105 ± 3kms−1 that the (unidentified) DIB carriers may only exist in the diffuse (Ca ii 849.8 nm), and is centered at the rest-frame velocity of medium, not in the dark cloud environment of M 17. the star. The Ca ii triplet lines are probably produced in an op- tically thick medium, since their strength ratio is not 1:9:5. The strongest lines of the H i Paschen and Brackett series also ex- 3.3. Spectral energy distribution hibit a central emission component, though it is single-peaked. Figure 3 shows the flux-calibrated X-shooter spectrum (300− The higher series members include a weaker emission compo- 2500 nm) of B275. The photometric data points demonstrate nent that may be double-peaked. A number of metallic emission the accuracy of the spectrophotometric calibration. The long lines (e.g., C i and Fe ii) are detected throughout the spectrum, standing debate over the distance to M 17 (ranging from 1.3 albeit very weak. to 2.1 kpc) has recently been settled by the measurement of the Prominent CO 1st-overtone emission bandheads are detected trigonometric of the CH3OH maser source G15.03–0.68 at 2.3 μm, with clear evidence of a blue shoulder. We also con- +0.14 (Xu et al. 2011), resulting in a distance of 1.98−0.12 kpc so that firm the presence of 2nd-overtone CO bandhead emission at M 17 is likely located in the Carina-Sagittarius spiral arm. 1.5 μm(Hanson et al. 1997). CO is easily dissociated so must We deredden the flux-calibrated X-shooter spectrum of B275 be shielded from the strong UV flux of the young massive star. (Fig. 3) using the parameterization of the extinction law by On the other hand, to produce 1st overtone emission, CO must Cardelli et al. (1989). The dereddened spectrum is fit to a Kurucz be excited, requiring a temperature in the range between 1500 model (Kurucz 1979, 1993) based on an iterative procedure, with and 4500 K (Bik & Thi 2004). This temperature might even fixed parameters Teff = 13 000 K, log g = 3.5, d = 1.98 kpc and be higher, considering the unprecedented detection of 2nd over- RV = 3.3(aneffective value resulting from interstellar and local tone emission. These conditions can be met in the plane of a extinction). This yields independent best-fit values of AV = 6.1 ± dense circumstellar disk where the CO molecules can be formed, 0.6 mag and R = 8.1 ± 0.8 R. Note that this radius is much excited, and protected from dissociation through self-shielding.

L1, page 3 of 4 A&A 536, L1 (2011)

6 × 5 1e+4 yr PMS objects are less than 5 10 yr old. Based on its location 3e+4 yr on the HRD we estimate the age of B275 at 105 yr. 1e+5 yr B275 bears some resemblance to a classical . 3e+5 yr 5 However, we note that Be stars do not emit CO 1st and 2nd over- tone emission. In addition, B275 would be classified as a lu- minous Herbig Be star according to the definition discussed in 12 Carmona et al. (2010), but our analysis of the photospheric spec- 4 10 trum allows for a more quantitative classification. B275 has a significant amount of infrared excess, starting at 8 1 μm, and a flat SED between 2 and 10 μm(Nielbock et al.

Log L/L 3 2001). This indicates the presence of a flaring circumstellar 6 disk in which the dust has not settled yet. However, the visi- bility of the photosphere, the small number of optical gas emis- sion lines and the absence of a jet lead us to believe that the 2 current mass-accretion rate is not very high. Nevertheless, the CO (2−0) and (3−0) emission originates in a dense and highly excited inner part of the disk. Either the system is in an intermit- tent phase between accretion episodes, or it is on the verge of 1 4.6 4.4 4.2 4 3.8 3.6 3.4 photo-evaporating its disk. Either scenario is consistent with its Log T location in the HR-diagram: an intermediate-mass, visible star eff in M 17 on its way to becoming an early-B main-sequence star. Fig. 4. The location of B275 (red parallellogram) in the HRD next to PMS tracks from Hosokawa & Omukai (2009) with the ZAMS mass labeled and open symbols indicating lifetimes. The thin dashed and thin −4 −1 dot-dashed lines are the birth lines for accretion rates of 10 M yr Acknowledgements. We thank Takashi Hosokawa for kindly providing the PMS −5 −1 ff and 10 M yr , respectively; the thick solid line is the ZAMS tracks. The ESO Paranal sta is acknowledged for obtaining the X-shooter (Schaller et al. 1992). The filled and open circles represent stars in M 17 spectrum of B275. We thank the anonymous referee for useful comments and suggestions. for which a spectral type has been determined (Hoffmeister et al. 2008), within a radius of 0.5and1.0, respectively; dots are other stars in M 17. B275 is on its way to becoming a 6−8 M ZAMS star. References Appenzeller, I., & Mundt, R. 1989, A&ARv, 1, 291 Benjamin, R. A., Churchwell, E., Babler, B. L., et al. 2003, PASP, 115, 953 The relative strength and shape of the CO bandheads can be Bik, A., & Thi, W. F. 2004, A&A, 427, L13 Bik, A., Kaper, L., & Waters, L. B. F. M. 2006, A&A, 455, 561 modeled by an optically thin Keplerian disk (Bik & Thi 2004; Blum, R. D., Barbosa, C. L., Damineli, A., Conti, P. S., & Ridgway, S. 2004, Blum et al. 2004), where the blue shoulder would imply a rela- ApJ, 617, 1167 tively high inclination angle of the disk (“edge-on”). Blum et al. Broos, P. S., Feigelson, E. D., Townsley, L. K., et al. 2007, ApJS, 169, 353 (2004) model the CO 2−0 first-overtone ro-vibrational bandhead Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245 at 2294 nm of B275 resulting in v sin i = 109.7 ± 0.6kms−1 Carmona, A., van den Ancker, M. E., Audard, M., et al. 2010, A&A, 517, A67 Chini, R., Elsaesser, H., & Neckel, T. 1980, A&A, 91, 186 (at the inner edge of the CO emission zone) and surface Chini, R., Hoffmeister, V., Kimeswenger, S., et al. 2004, Nature, 429, 155 21 −2 NCO = 3.5 ± 0.2 × 10 cm . The double-peaked emission pro- Cox, N. L. J., Kaper, L., Foing, B. H., & Ehrenfreund, P. 2005, A&A, 438, 187 files, as shown in Fig. 1, are very similar to the emission-line D’Odorico, S., Dekker, H., Mazzoleni, R., et al. 2006, in SPIE Conf. Ser., 6269 profile of a single line obtained by Blum et al. (2004). Ellerbroek, L. E., Kaper, L., Bik, A., et al. 2011, ApJ, 732, L9 Goldoni, P., Royer, F., François, P., et al. 2006, in SPIE Conf. Ser., 6269 We find no evidence for veiling of the optical spectrum or Gray, R. O., & Corbally, J., C. 2009, Stellar Spectral Classification (Princeton any strong indications of active “heavy” accretion and/or jets, University Press) such as those observed in some other systems (e.g., Ellerbroek Hanson, M. M., Howarth, I. D., & Conti, P. S. 1997, ApJ, 489, 698 et al. 2011). The [O ii] 630 nm line very likely has a nebular Hanson, M. M., Luhman, K. L., & Rieke, G. H. 2002, ApJS, 138, 35 Hoffmeister, V. H., Chini, R., Scheyda, C. M., et al. 2008, ApJ, 686, 310 origin. Hosokawa, T., & Omukai, K. 2009, ApJ, 691, 823 Hosokawa, T., Yorke, H. W., & Omukai, K. 2010, ApJ, 721, 478 Keto, E., Broderick, A. E., Lada, C. J., & Narayan, R. 2006, ApJ, 652, 1366 Kraus, S., Hofmann, K., Menten, K. M., et al. 2010, Nature, 466, 339 4. Discussion Krumholz, M. R., Klein, R. I., McKee, C. F., Offner, S. S. R., & Cunningham, A. J. 2009, Science, 323, 754 The accurate spectral classification and SED fit result in a well- Kurucz, R. L. 1979, ApJS, 40, 1 Kurucz, R. L. 1993, VizieR Online Data Catalog, 6039, 0 defined position of B275 in the Hertzsprung-Russell diagram Modigliani, A., Goldoni, P., Royer, F., et al. 2010, in SPIE Conf. Ser., 7737 (HRD, Fig. 4). It is located well above the ZAMS, demonstrat- Nielbock, M., Chini, R., Jütte, M., & Manthey, E. 2001, A&A, 377, 273 ing its PMS nature. If B275 is contracting towards the ZAMS, Povich, M. S., Churchwell, E., Bieging, J. H., et al. 2009, ApJ, 696, 1278 the final ZAMS mass would be 6−8 M (spectral type B1−B2), Puls, J., Urbaneja, M. A., Venero, R., et al. 2005, A&A, 435, 669 assuming that no additional mass is accreted. Schaller, G., Schaerer, D., Meynet, G., & Maeder, A. 1992, A&AS, 96, 269 Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163 To be visible at this location in the HRD, the star must have Testi, L., Tan, J. C., & Palla, F. 2010, A&A, 522, A44 −5 −1 experienced an average accretion rate of at least 10 M yr Urquhart, J. S., Moore, T. J. T., Hoare, M. G., et al. 2011, MNRAS, 410, 1237 in its recent history. B275 may thus be the long-sought-for ex- Vernet, J., Dekker, H., D’Odorico, S., et al. 2011, A&A, in press, DOI: 10.1051/0004-6361/201117752 ample of an early-B PMS star. Figure 4 also shows other nearby Wheelwright, H. E., Oudmaijer, R. D., de Wit, W. J., et al. 2010, MNRAS, 408, stars in M 17. Hanson et al. (1997) derive an age of ∼1Myrfor 1840 the M 17 cluster andHoffmeister et al. (2008) estimate that the Xu, Y., Moscadelli, L., Reid, M. J., et al. 2011, ApJ, 733, 25

L1, page 4 of 4